Activity and Cytosolic Na⁺ Regulate Synaptic Vesicle Endocytosis

Abstract

Retrieval of synaptic vesicles via endocytosis is essential for maintaining sustained synaptic transmission, especially for neurons that fire action potentials at high frequencies. However, how neuronal activity regulates synaptic vesicle recycling is largely unknown. Here we report that Na⁺ substantially accumulated in the mouse calyx of Held terminals of either sex during repetitive high-frequency spiking. Elevated presynaptic Na⁺ accelerated both slow and rapid forms of endocytosis and facilitated endocytosis overshoot, but did not affect the readily releasable pool size, Ca²⁺ influx, or exocytosis. To examine whether this facilitation of endocytosis is related to the Na⁺-dependent vesicular content change, we dialyzed glutamate into the presynaptic cytosol or blocked the vesicular glutamate uptake with bafilomycin and found that the rate of endocytosis was not affected by regulating the vesicular glutamate content. Endocytosis is critically dependent on intracellular Ca²⁺, and the activity of Na⁺/Ca²⁺ exchanger (NCX) may be altered when the Na⁺ gradient is changed. However, neither NCX inhibitor nor change of extracellular Na⁺ concentration affected the endocytosis rate. Moreover, two-photon Ca²⁺ imaging showed that presynaptic Na⁺ did not affect the action potential-evoked intracellular Ca²⁺ transient and decay. Therefore, we revealed a novel mechanism of cytosolic Na⁺ in accelerating vesicle endocytosis. During high-frequency synaptic transmission, when large numbers of synaptic vesicles were fused, the rapid buildup of presynaptic cytosolic Na⁺ promoted vesicle recycling and sustained synaptic transmission.

SIGNIFICANCE STATEMENT High-frequency firing neurons are widely distributed in the CNS. A large number of synaptic vesicles are released during high-frequency synaptic transmission; accordingly, synaptic vesicles need to be recycled rapidly to replenish the vesicle pool. Synaptic vesicle exocytosis and endocytosis are tightly coupled, and their coupling is essential for synaptic function and structural stability. We showed here that intracellular Na⁺ concentration at the calyx of Held terminal increased rapidly during spike activity and the increased Na⁺ accelerated endocytosis. Thus, when large numbers of synaptic vesicles are released during high-frequency synaptic transmission, Na⁺ accumulated in terminals and facilitated vesicle recycling. These findings represent a novel cellular mechanism that supports reliable synaptic transmission at high frequency in the CNS.

Introduction

At chemical synapses, the fusion of synaptic vesicles with the presynaptic plasma membrane releases vesicular neurotransmitter contents for exerting various functions. Following exocytosis, membrane lipid bilayer and proteins are retrieved through endocytosis to form new vesicles (Sudhof, 2004; Kononenko and Haucke, 2015; Soykan et al., 2017). Different modes of endocytosis have been described in central synapses, including the classical clathrin-dependent endocytosis, kiss-and-run, and bulk endocytosis (Hosoi et al., 2009; Wu et al., 2009; Yamashita et al., 2010). Endocytosis in different conditions differs substantially in speed and amount. Slow endocytosis is mediated by a classical, clathrin- and dynamin-dependent endocytosis with a decay time constant (τ) of ∼10 to 30 s, serving as the predominant mode for synaptic vesicle recycling during low-intensity activity (Granseth et al., 2006). The clathrin-independent, dynamin-dependent rapid endocytosis, with a τ within a few seconds, has been assumed to reflect kiss-and-run that involves rapid fusion pore opening and closure of the same vesicles after stronger stimulation (Alés et al., 1999; Wu et al., 2005). Bulk endocytosis occurs when large endosome-like structures are internalized from presynaptic plasma membrane during high-intensity firing activity (Holt et al., 2003; Wu and Wu, 2007). Moreover, during intense stimulation and a massive influx of Ca²⁺, the synaptic terminals may retrieve more membranes than vesicles being exocytosed. This phenomenon is defined as endocytosis overshoot that has been implicated in increasing endocytosis capacity and efficiency during high-frequency firing (Renden and von Gersdorff, 2007; Wu et al., 2009; Xue et al., 2012).

Accumulating evidence indicated that synaptic vesicle exocytosis and endocytosis are tightly coupled both temporally and spatially, and their timely coupling is essential for synaptic function and structural stability, but the underlying mechanism of this coupling is under debate. Ca²⁺ represents one prevalent possibility. Vesicle exocytosis is directly triggered by Ca²⁺, while how Ca²⁺ regulates endocytosis is intricate and varies greatly in response to distinct neuronal activity (Leitz and Kavalali, 2016), making it complicated to coordinate endocytosis speed in an activity-dependent manner. Here we showed that action potential firing substantially elevated presynaptic cytosolic Na⁺ in an activity-dependent manner in the mouse calyx of Held, a giantglutamatergic terminal in the auditory brainstem that fires continuous action potentials at high frequency up to hundreds of hertz. We found that the cytosolic Na⁺ accelerated both slow and rapid forms of endocytosis and facilitated endocytosis overshoot, while the Na⁺ effect on endocytosis was not implemented either through affecting Ca²⁺ influx or intracellular Ca²⁺ transients, or through regulating vesicular glutamate contents. Therefore, when large numbers of synaptic vesicles were fused during high-frequency synaptic transmission, Na⁺ accumulated into presynaptic cytosol and facilitated vesicle recycling. Since thecytosolic Na⁺ concentration is correlated with spike intensity and duration, it may work as a signal to coordinate vesicle endocytosis and recycling according to the level of exocytosis.

Materials and Methods

Slice preparation.

The care and handling of animals were approved by the Institutional Animal Care and Use Committee of Tulane University and complied with US Public Health Service guidelines. Coronal brainstem slices containing the medial nucleus of the trapezoid body were prepared from postnatal day 8–12 C57BL/6J mice of either sex similar to those previously described (Zhang and Huang, 2017). Briefly, 210 μm slices were cut using a Vibratome (VT1200S, Leica) in ice-cold, low-Ca²⁺, low-Na⁺ saline containing the following (in mm): 230 sucrose, 10–25 glucose, 2.5 KCl, 0.1 CaCl₂, 3 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, 0.4 ascorbic acid, 3 myo-inositol, and 2 Na-pyruvate, bubbled with 95% O₂/5% CO₂. Slices were incubated at 32°C for 20–30 min and thereafter stored at room temperature in normal artificial CSF (aCSF) contained the following (in mm): 125 NaCl, 10–25 glucose, 2.5 KCl, 1.2 CaCl₂, 1.8 MgCl₂, 1.25 NaH₂PO₄, 25 NaHCO₃, 0.4 ascorbic acid, 3 myo-inositol, and 2 Na-pyruvate, pH 7.4, bubbled with 5% CO₂/95% O₂ before use.

Membrane capacitance measurements.

Slices were transferred to a recording chamber and perfused with bubbled aCSF (2–3 ml/min) warmed to ∼32°C by an inline heater (Warner Instruments). Whole-cell recordings were made from calyces of Held with an EPC-10 USB patch-clamp amplifier with lock-in system and PatchMaster software (HEKA). For whole-cell membrane capacitance (C_m) measurements, the sinusoidal stimulus frequency was 1 kHz and the peak-to-peak voltage was 60 mV (Lindau and Neher, 1988). To isolate presynaptic Ca²⁺ currents in voltage-clamp experiments, TEA-Cl (10 mm), 4-aminopyridine (0.5 mm), and tetrodotoxin (1 μm) were added to aCSF, substituting for NaCl with equal osmolarity. For recordings of the endocytosis overshoot, MgCl₂ and CaCl₂ were adjusted to 1.0 and 2.0 mm, respectively, to facilitate overshoot (Wu et al., 2009; Xue et al., 2012). To test endocytosis under lowered extracellular Na⁺ concentration, 50 mm NaCl was replaced by LiCl with equal osmolarity (Kim et al., 2005). The pipette solution contained the following (in mm): 70 Cs-methanesulfonate, 20 CsCl, 10 HEPES, 0.5 EGTA, 4 Mg-ATP, 0.3 Tris₃-GTP, 10 Tris₂-phosphocreatine, 5 glutamate, as well as 40 NMDG-methanesulfonate (for Na⁺-free solution), 10 Na-methanesulfonate + 30 NMDG-methanesulfonate (for 10 mm Na⁺ solution), or 40 Na-methanesulfonate (for 40 mm Na⁺ solution). For presynaptic glutamate dialysis experiments, glutamate was added to substitute methanesulfonate with equal osmolarity. All solutions were adjusted to pH 7.3 with CsOH (310–315 mOsm). Tips of patch pipettes (3–5 MΩ) were coated with dental wax to reduce stray capacitance. Series resistance (<20 MΩ) was compensated up to 60% (10 μs lag). Data were obtained 4–20 min after break-in at sampling rate of 100 kHz and filtered with an online Bessel filter at 2.9 kHz.

Two-photon Na⁺ and Ca²⁺ imaging.

A Galvo multiphoton microscopy system (Scientifica) with a Ti:sapphire pulsed laser (Chameleon Ultra II, Coherent) was used for Na⁺ and Ca²⁺ imaging. Whole-cell patch-clamp recordings were performed with a Multiclamp 700B amplifier and pClamp software (Molecular Devices). The laser was tuned to 800 nm for Na⁺ imaging and 810 nm for Ca²⁺ imaging, and epifluorescence signals were captured through 60×, 1.0 numerical aperture (NA) objectives and a 1.4 NA oil-immersion condenser (Olympus). Fluorescence was split into red and green channels using dichroic mirrors and bandpass filters. Data were collected in frame-scan or line-scan modes using SciScan (Scientifica) or ScanImage (Vidrio Technologies). Presynaptic spikes were evoked by afferent fiber stimulation or current injection and recorded under current clamp, and corresponding Ca²⁺ and Na⁺ signals were recorded under two-photon imaging.

For Ca²⁺ imaging under different Na⁺ concentrations, the pipette solution contained the following (in mm): 70 K-methanesulfonate, 20 KCl, 10 HEPES, 4 Mg-ATP, 0.3 Tris₃-GTP, 10 Tris₂-phosphocreatine, 5 glutamate, as well as 40 NMDG-methanesulfonate (for Na⁺-free solution), 10 Na-methanesulfonate + 30 NMDG-methanesulfonate (for 10 mm Na⁺ solution), or 40 Na-methanesulfonate (for 40 mm Na⁺ solution). For Ca²⁺ imaging with bath application of KB-R7943, the pipette solution contained the following (in mm): 130 K-gluconate, 20 KCl, 4 MgATP, 0.3 Tris-GTP, 7 Na₂-phosphocreatine, and 10 HEPES. All solutions were adjusted to 290 mOsm, pH 7.3 with KOH; 250 μm Fluo-5F and 20 μm Alexa Fluor 594 were also added to the pipette solutions before the experiments. Data are expressed as Δ(G/R)/(G/R)_max × 100%, where (G/R)_max is the maximal fluorescence in saturating Ca²⁺ (Spratt et al., 2019).

For Na⁺ imaging, pipette solution contained the following (in mm): 110 K-methanesulfonate, 20 KCl, 10 HEPES, 0.5 EGTA, 4 Mg-ATP, 0.3 Tris₃-GTP, and 5 Na₂-phosphocreatine, 290 mOsm and pH 7.3 with KOH, while 1 mm sodium-binding benzofurane isophthalate (SBFI) and 15 μm Alexa Fluor 594 were added before the experiments. To avoid the perfusion of cytoplasmic Na⁺ during resting and spike-evoked Na⁺ signals recordings, pipettes were subsequently detached after whole-cell dialysis with SBFI and Alexa Fluor 594 dyes when the fluorescence intensities were stable and Na⁺ signals were measured after waiting at least 10 min. Standard calibration methods were used to measure absolute Na⁺ concentrations (Rose, 2012; Huang and Trussell, 2014). After loading fluorescent dyes through whole-cell recordings, pipettes were detached for resting [Na⁺] and calibration. The solutions for in situ calibration of SBFI fluorescence contained the following (in mm): 20 KCl,25 glucose, 10 HEPES, and 130 (K-gluconate + Na-gluconate), and adjusted to pH 7.4 with KOH; 3 μm gramicidin D, 10 μm monensin,and 50 μm ouabain were added into the calibration solutions before experiments.

Drugs.

Drugs were obtained from Alomone Labs (tetrodotoxin), LC Laboratories (bafilomycin A1), and Thermo Fisher Scientific (SBFI, fluo-5F, and Alexa Fluor 594), and all others were obtained from Sigma-Aldrich. Bafilomycin A1 and KB-R7943 were prepared with DMSO, stored at –20°C, and dissolved in aCSF immediately before experiments.

Analysis.

Data were analyzed using PatchMaster (HEKA), Clampfit (Molecular Devices), Igor (WaveMetrics), and ImageJ (NIH). Liquid junction potentials (10–11 mV) were measured and adjusted appropriately. The following standard monoexponential functions were used to describe slow endocytosis: f(t) = A × e^−t/τ + C, where A is the capacitance jump (ΔC_m) and τ is the time constant. Double exponential functions were used to describe endocytosis under strong stimulations, as follows: f(t) = A₁ × e^−t/τ1 + A₂ × e^−t/τ2 + C, where A₁ and A₂ are the amplitudes of fast and slow exponential components; and τ₁ and τ₂ are the time constants of each component, respectively. The weighted time constant is expressed: τ_weight = (A₁ × τ₁ + A₂ × τ₂)/(A₁ + A₂). For Na⁺ calibration, data are normalized and fitted by the following equation: (ΔG/R) = (G/R)_max × [Na⁺]_i/([Na⁺]_i + K_app), where G/R is the ratio of green fluorescence relative to red fluorescence; (ΔG/R) is the change in fluorescence ratio measured at a given [Na⁺]_i divided by that at 0 mm [Na⁺]_i; (G/R)_max is the maximal change in fluorescence ratio and K_app is the apparent K_d of SBFI. Data were presented as the mean ± SEM. Statistical significance was established using one-way ANOVA followed by Tukey's multiple-comparisons test, as well as paired and unpaired t tests as indicated, with p < 0.05 indicating a significant difference.

Results

Spikes activity increases presynaptic cytosolic Na⁺ concentration

Changes in presynaptic Na⁺ during action potential (AP) firing were assayed using Na⁺ imaging with two-photon laser scanning microscopy. Calyces of Held were loaded via whole-cell recordings with the Na⁺ indicator SBFI and the volume marker Alexa Fluor 594. Standard calibration methods (Rose, 2012; Huang and Trussell, 2014) were used to measure the absolute [Na⁺] (Fig. 1A). The presynaptic [Na⁺] at the resting state was 15.8 ± 2.1 mm (n = 4). APs were evoked by afferent fiber stimulation and propagated to the presynaptic terminals. Upon 100 Hz stimulation for 1 s, presynaptic cytosolic [Na⁺] increased by 5.7 ± 1.1 mm, which decayed to the control level with a time constant of 12.2 ± 1.0 s (n = 5; Fig. 1B). After ensuring reliable AP evocation by afferent fiber stimulation, the recording pipettes were subsequently detached from the calyces, and the Na⁺ signals on stimulations at different frequencies were measured. Spiking at 10 Hz for 60 s reversibly increased the [Na⁺] by 12.2 ± 1.7 mm (n = 7; Fig. 1C). Na⁺ transients were gradually augmented as increasing spike frequency, as follows: 20 Hz for 30 s increased the [Na⁺] by 16.3 ± 2.0 mm (n = 8; Fig. 1D) and 100 Hz for 20 s increased the [Na⁺] by 55.6 ± 5.9 mm (n = 8; Fig. 1E). Therefore, spike activities efficiently increase the presynaptic cytosolic Na⁺ concentration in an activity-dependent manner (Fig. 1F).

Figure 1.

Presynaptic spikes control cytosolic Na⁺ concentration. A, In situ calibration of SBFI fluorescence under two-photon microscopy. Top, Single optical section of a calyx of Held terminal filled with SBFI (green) and Alexa Fluor 594 (red) through whole-cell recording. The recording pipette was subsequently detached for resting [Na⁺] measurement and [Na⁺] calibration. Bottom, Change in fluorescence with [Na⁺]_i (n = 5). The fitted curve yielded a K_app of 25.7 mm and (G/R)_max of –0.86 for SBFI (see Materials and Methods). The resting [Na⁺]_i was shown in the insert. B, A calyx of Held filled with SBFI and Alexa Fluor 594 under whole-cell recording. Spikes were evoked at 100 Hz for 1 s by afferent fiber stimulation. An increase in [Na⁺] was visualized under line-scan and frame-scan modes. C–E, After detaching recording pipettes from calyces, Na⁺ signals in response to increasing spike frequencies were measured. Spikes at 10 Hz (C), 20 Hz (D), and 100 Hz (E) induced different changes in intracellular Na⁺ concentration. F, Summary of results of presynaptic Na⁺ increases under different stimulations. Error bars indicate ±SEM.

Presynaptic cytosolic Na⁺ facilitates slow endocytosis

Capacitance measurements were made at the calyx of Held terminals to examine whether presynaptic Na⁺ influences vesicle endocytosis and recycling. Endocytosis of the calyces depends on the intensity of activity. Mild stimulations such as single pulses of 1–40 ms depolarization trigger slow clathrin-dependent endocytosis, while stronger stimulations such as 10 pulses of 20 ms depolarization induce an additional clathrin-independent rapid form of endocytosis (Wu et al., 2005, 2009; Hosoi et al., 2009; Yamashita et al., 2010). We measured the whole-cell capacitance under 1.2 mm extracellular Ca²⁺ at 32°C with pipette solutions containing 0, 10, or 40 mm Na⁺. A voltage depolarization from –80 to +10 mV for 40 ms (depol_40ms) induced Ca²⁺ influx and triggered ΔC_m, followed by slow C_m decay toward to the prestimulus level (Fig. 2). Because C_m is proportional to the membrane surface area, ΔC_m reflects exocytosis and C_m decay reflects endocytosis (Sun and Wu, 2001). This protocol is sufficient to release the readily releasable vesicles (Wu and Borst, 1999; Fedchyshyn and Wang, 2005; Renden and von Gersdorff, 2007), with ΔC_m reflecting the size of readily releasable pool. We found that the Ca²⁺ currents (p = 0.87) and ΔC_m (p = 0.47; one-way ANOVA test) were not different in recordings with 0, 10, or 40 mm Na⁺ (Fig. 2C,D), while presynaptic Na⁺ showed profound effects on the endocytosis rate (Fig. 2E). When the terminal was dialyzed with 10 mm Na⁺, the C_m decay could be fitted with a monoexponential function with a time constant (τ) of 15.3 ± 0.9 s (n = 9). With the 40 mm Na⁺ solution dialyzed presynaptically, the endocytosis rate accelerated, and the time constant reduced to 10.4 ± 1.0 s (n = 10). Upon dialysis with Na⁺-free solution, the endocytosis became slower, resulting in a time constant of 25.1 ± 1.9 s (n = 9; p < 0.0001, one-way ANOVA followed by Tukey's multiple-comparisons test). Thus, presynaptic cytosolic Na⁺ accelerates slow endocytosis induced by single depolarizing steps.

Figure 2.

Presynaptic Na⁺ facilitates slow endocytosis without affecting exocytosis. A, Sampled presynaptic Ca²⁺ currents induced by 40 ms depolarizations from –80 to +10 mV (depol_40ms) with presynaptic pipette solutions containing 0 mm (left), 10 mm (middle), or 40 mm Na⁺ (right). B, Corresponding C_m recordings showing exocytosis and endocytosis induced by depol_40ms. Membrane conductance (G_m) and series conductance (G_s) are included to confirm the recording stability. C–E, Group data of the effects of cytosolic Na⁺ on Ca²⁺ charge (Q_Ca; C), capacitance jump (ΔC_m; D), and the time constant of capacitance decay (τ; E). *p < 0.05, ***p < 0.001, ****p < 0.0001; Tukey's multiple-comparisons test. Error bars indicate ±SEM.

Cytosolic Na⁺ facilitates rapid endocytosis

Stronger stimulation induces a clathrin-independent rapid form of endocytosis along with the slow endocytosis (Alés et al., 1999; Wu et al., 2005), so we next investigated whether Na⁺ modulates the rapid form of endocytosis. With 10 mm Na⁺ in the presynaptic solution, 10 pulses of 20 ms depolarization from −80 to +10 mV at 10 Hz (depol_20ms × 10) evoked a ΔC_m of 1002 ± 55 followed by a C_m decay that can be fitted with a double exponential function whose fast and slow time constants were 3.3 ± 0.3 and 15.1 ± 1. 9 s, respectively (Fig. 3). The fast component was 40.1 ± 2.7% of the fit (the remainder being slow component), and the weighted mean was 11.5 ± 1.1 s (n = 11). When the terminal was dialyzed with 40 mm Na⁺, both fast and slow endocytosis were accelerated, resulting in fast and slow components of 1.3 ± 0.2 s (36.4 ± 3.4%) and 10. 8 ± 1.1 s, respectively, and a weighted time constant of 7.6 ± 0.8 s (n = 11), while Na⁺-free solution slowed down both fast and slow endocytosis to 4.9 ± 0.6 s (36. 6 ± 4.7%) and 22.6 ± 2.1 s, respectively, and a weighted time constant of 18.3 ± 1.8 s (n = 10; p < 0.0001, one-way ANOVA). The evoked ΔC_m was not different among groups (Fig. 3E; p = 0.67, one-way ANOVA). These results revealed that presynaptic cytosolic Na⁺ facilitates both the slow and rapid modes of synaptic vesicle endocytosis induced by repetitive depolarization steps.

Figure 3.

Cytosolic Na⁺ facilitates rapid endocytosis. A–C, Sampled C_m recordings showing exocytosis and endocytosis induced by 10 of 20 ms depolarizations (depol_20ms × 10) with pipette solutions containing 0 mm (A), 10 mm (B), or 40 mm (C) Na⁺. D, Normalized C_m traces of A–C showing the Na⁺ effects on endocytosis rate. E, Statistics for ΔC_m. F–H, The capacitance decay was fitted by double exponentials. The fast (τ_fast; F) and slow (τ_slow; G) components, and the weighted mean (τ_weight; H) from different cytosolic Na⁺ concentrations were compared. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; Tukey's multiple-comparisons test. Error bars indicate ±SEM.

Cytosolic Na⁺ facilitates endocytosis overshoot

During intense stimulations, such as 10 depolarization pulses of 50 ms at 10 Hz (depol_50ms × 10), the calyx terminal could retrieve more membranes than exocytosed, and induced endocytosis overshoot (Renden and von Gersdorff, 2007; Wu et al., 2009; Xue et al., 2012). We then tested whether Na⁺ influences the endocytosis overshoot, in which an extracellular solution containing 2 mm Ca²⁺ was used to increase the chance of the observation (Wu et al., 2009; Xue et al., 2012). The size of endocytosis overshoot was quantified as the capacitance value below the baseline at 40∼60 s after depolarization pulses (Wu et al., 2009). We found that increased Na⁺ concentrations facilitated the observation of endocytosis overshoot. As the [Na⁺] increased from 0 to 10, and 40 mm, the percentage of cells with overshoot increased from 28.6% (two of seven calyces) to 66.7% (six of nine calyces), and 100% (seven of seven calyces). For the cells that showed overshoot, the overshoot ratio increased from 5.5 ± 4.4% to 14.7 ± 3.0%, and 22.8 ± 4.3% (Fig. 4; p = 0.0008, one-way ANOVA). These results indicate that the cytosolic Na⁺ also facilitates endocytosis overshoot in both the percentage of cells and the size of the membrane area.

Figure 4.

Cytosolic Na⁺ facilitates endocytosis overshoot. A, Sampled endocytosis overshoots induced by 10 50 ms depolarizations (depol_50ms × 10) with presynaptic pipette solutions containing 0, 10, or 40 mm Na⁺. Extracellular Ca²⁺ was elevated to 2 mm to increase the chance of observing overshoot. B, Statistics for endocytosis overshoot size in the three groups. *p < 0.05, ***p < 0.001; Tukey's multiple-comparisons test. Error bars indicate ±SEM.

Vesicular content does not affect slow or rapid endocytosis

A previous study showed that cytosolic transmitter concentration, which rapidly controls the vesicular neurotransmitter contents (Hori and Takahashi, 2012; Apostolides and Trussell, 2013), regulates vesicle cycling at hippocampal GABAergic terminals in culture (Wang et al., 2013). Since presynaptic Na⁺ facilitates vesicular glutamate transport and higher cytosolic [Na⁺] increases the vesicle glutamate contents (Huang and Trussell, 2014; Li et al., 2020), we performed two lines of experiment to test whether the effects of Na⁺ on endocytosis occur through regulating vesicular contents (Fig. 5). First, we performed capacitance measurements under different concentrations of cytosolic glutamate (0, 5, or 50 mm), which allows the manipulation of vesicular glutamate content (Hori and Takahashi, 2012), while keeping the cytosolic [Na⁺] constant at 10 mm. Slow and rapid endocytosis were evoked by the weak (depol_40ms; Fig. 5A) and intense (depol_20ms × 10; Fig. 5E) stimulations, respectively. Our results showed that neither the slow (p = 0.82, one-way ANOVA; Fig. 5D) nor the rapid (p = 0.74, one-way ANOVA; Fig. 5H) mode of endocytosis was influenced by the cytosolic glutamate concentration. Second, we tested the effects of bafilomycin, a V-ATPase inhibitor that impedes vesicle acidification and neurotransmitter reloading, on endocytosis triggered by the same protocols (Fig. 5B,F). The results showed that 2 μm bafilomycin affected neither slow (p = 0.30, unpaired t test; Fig. 5D) nor rapid endocytosis (p = 0.56, unpaired t test; Fig. 5H). The exocytoses were not different among all groups (Fig. 5C,G). These results indicate that vesicular content had no significant effects on exocytosis or endocytosis.

Figure 5.

Vesicular glutamate content does not affect the endocytosis rate. A, B, Slow endocytosis induced by depol_40ms when pipette solutions contained 0, 5, or 50 mm glutamate (A), or after incubating bafilomycin A1 (B). C, D, Neither glutamate nor bafilomycin affected the capacitance jump ΔC_m (C) or the time constant of capacitance decay τ (D). E–H, Similar to A–D, while rapid endocytosis was induced by depol_20ms × 10. Error bars indicate ±SEM.

Cytosolic Na⁺ facilitated endocytosis is independent of NCX activity

It has been shown that plasma membrane Na⁺/Ca²⁺ exchanger (NCX) is involved in maintaining presynaptic Ca²⁺ homeostasis (Kim et al., 2005), while Ca²⁺ plays specific roles in synaptic vesicle endocytosis (Leitz and Kavalali, 2016). To test whether presynaptic cytosolic Na⁺ affects endocytosis by regulating NCX function, we lowered the extracellular Na⁺ concentration to mimic the Na⁺ concentration gradient change in elevated intracellular Na⁺ conditions. As shown in Figure 6A–C, the endocytosis rate was not changed by lowering extracellular Na⁺ for 50 mm (p = 0.50, n = 6, paired t test). We also tested whether inhibiting NCX activity directly affects the endocytosis rate. As shown in Figure 6D–F, incubation with the NCX inhibitor KB-R7943 (20 μm) slightly but not significantly slowed the rate of endocytosis (n = 8; p = 0.14, paired t test). To confirm that KB-R7943 indeed inhibited Na⁺/Ca²⁺ exchange activity, we performed two-photon Ca²⁺ imaging with Fluo-5F. Ca²⁺ influx was evoked by 10 APs at 100 Hz. KB-R7943 did not affect the peak Ca²⁺ signal (p = 0.26, n = 5), while the decay time constant was increased from 0.57 ± 0.07 to 0.84 ± 0.13 s (Fig. 6G–I; n = 5; p = 0.02, paired t test). On the other hand, if Na⁺ regulates endocytosis through slowing NCX-dependent Ca²⁺ extrusion, one would, however, expect an acceleration of endocytosis by NCX inhibitor. Therefore, the presynaptic Na⁺-regulated vesicle recycling is not mediated by modulating NCX function or Ca²⁺ extrusion.

Figure 6.

Facilitation of endocytosis by Na⁺ does not occur through affecting Na⁺/Ca²⁺ exchange activity. A, D, Sampled C_m recordings showing exocytosis and endocytosis induced by 10 20 ms depolarizations (depol_20ms × 10) in control and with low extracellular Na⁺ concentration (A) or in the presence of NCX inhibitor KB-R7943 (D). B, C, Low extracellular Na⁺ did not affect the ΔC_m (B) or the weighted time constant of capacitance decay (τ_weight; C). E, F, KB-R7943 did not affect ΔC_m (E) or τ_weight (F). G–I, Two-photon Ca²⁺ imaging showing that the spike-evoked Ca²⁺ decay was slowed by KB-R7943 while the peak of Ca²⁺ transients stayed unchanged. *p < 0.05. Error bars indicate ±SEM.

Cytosolic Na⁺ does not affect spike-evoked intracellular Ca²⁺ rise and decay

To test whether cytosolic Na⁺ affects intracellular Ca²⁺ dynamics during spiking activity, we made two-photon Ca²⁺ imaging with Fluo-5F loaded into the presynaptic terminal (Fig. 7). Calyces were recorded with pipette solutions containing 0, 10, or 40 mm Na⁺. A burst of 10 APs at 100 Hz evoked a rapid Ca²⁺ rise of similar concentrations at all concentrations of Na⁺ tested, as indicated by the fluorescence increases of 29.3 ± 0.9% of (G/R)_max in Na⁺-free (n = 8), 30.3 ± 1.5% in 10 mm Na⁺ (n = 7), and 31.3 ± 1.6% in 40 mmNa⁺ (n = 6) pipette solutions (Fig. 7C). The Ca²⁺ signals decayed to the background level within seconds and were not different among groups. The Ca²⁺ decay time constants were 0.56 ± 0.06 s for 0 mm Na⁺, 0.60 ± 0.07 s for 10 mm Na⁺, and 0.61 ± 0.05 s for 40 mm Na⁺ solutions (p = 0.80, one-way ANOVA; Fig. 7D). Thus, it is unlikely that thecytosolic Na⁺ facilitates endocytosis through a Ca²⁺-dependent mechanism.

Figure 7.

Cytosolic Na⁺ does not affect spike-evoked Ca²⁺ transients and decay. A, Single optic section of the calyx with attached patch pipette. B, Presynaptic Ca²⁺ transients induced by 10 spikes at 100 Hz when dialyzed with a pipette solution containing 0, 10, or 40 mm Na⁺. C, D, Summary plots of relative Ca²⁺ rise and decay times with different intracellular Na⁺ concentrations. Error bars indicate ±SEM.

Discussion

Here we showed that intracellular Na⁺ concentration at the presynaptic terminal increased rapidly during spike activity (Fig. 1). This elevated [Na⁺] accelerated slow (Fig. 2) and rapid endocytosis (Fig. 3), and facilitated endocytosis overshoot (Fig. 4). The modulation of endocytosis by Na⁺ was unlikely related to vesicular glutamate contents (Fig. 5), Ca²⁺ influx through NCX (Fig. 6), or intracellular Ca²⁺ transients in response to stimulation (Fig. 7). Therefore, we revealed an unappreciated role of cytosolic Na⁺ in the regulation of synaptic vesicle endocytosis and recycling. When large numbers of synaptic vesicles are fused during high-frequency synaptic transmission, accumulated presynaptic Na⁺ accelerates vesicle recycling and sustains synaptic transmission, representing a novel cellular mechanism that supports reliable synaptic transmission at high- frequency in the CNS.

Accumulating evidence indicated that synaptic vesicle exocytosis and endocytosis are tightly coupled both temporally and spatially, and their coupling is essential for synaptic function and structural stability. During prolonged high-frequency synaptic transmission, when large numbers of synaptic vesicles are fused with presynaptic membrane, endocytosis needs to be accelerated to replenish the vesicle pool and maintain the strength of synaptic transmission. Slow endocytosis and vesicle recycling lead to depletion of the release-ready vesicles, which have been shown to be a major contribution to short-term depression (von Gersdorff and Borst, 2002; Fernández-Alfonso and Ryan, 2004; Regehr, 2012). Different mechanisms have been observed to couple endocytosis with exocytosis, including Ca²⁺ (Sankaranarayanan and Ryan, 2001; Hosoi et al., 2009; Wu et al., 2009), membrane lipids such as PIP2 (Koch and Holt, 2012), and cytoskeleton proteins (Yuan et al., 2015; Orlando et al., 2019). Ca²⁺-dependent regulation of exoendocytosis coupling has been prevalently studied. Exocytosis is directly triggered by Ca²⁺, and endocytosis is regulated by Ca²⁺-dependent mechanisms. (Wu et al., 2014). Ca²⁺ speeds up slow, rapid, and bulk endocytosis, while buffering intracellular Ca²⁺ with chelators or reducing Ca²⁺ influx slows down endocytosis in endocrine cells and various synapses (Sankaranarayanan and Ryan, 2001; Hosoi et al., 2009; Wu et al., 2009). However, increased Ca²⁺ influx during prolonged stimulation train slows down endocytosis in many preparations, including chromaffin cells, neuromuscular junction, retinalribbon terminals, and central synapses (von Gersdorff and Matthews, 1994; Wu and Betz, 1996; Sun et al., 2002, 2010; Elhamdani et al., 2006; Balaji et al., 2008). A recent study showed that the dynamics of Ca²⁺ concentration changes can differentially modulate endocytosis. Specifically, transient large calcium increases trigger endocytosis, while prolonged small global Ca²⁺ increases inhibit slow endocytosis (Wu and Wu, 2014). Therefore, the regulation of endocytosis by Ca²⁺ concentration is complex and varies greatly in response to distinct neuronal activity, making it complicated to coordinate endocytosis speed in an activity-dependent manner.

Neural activity is provided by ion fluxes through voltage and neurotransmitter-gated channels, and spiking activity alters ion composition in the cytosol. Although the intracellular K⁺ concentration is high and relatively stable, the Ca²⁺ and Na⁺ levels are dynamically regulated during activity. We showed that the presynaptic cytosolic Na⁺ of the mouse calyx of Held is 15.8 ± 2.1 mm, very similar to that of the rat calyx (Huang and Trussell, 2014). We previously reported that calyceal terminals express Na⁺-permeable HCN channels, and the activation of HCN channels contributes to the resting Na⁺ concentration of 4.9 ± 0.5 mm (Huang and Trussell, 2014). Here we found that spikes are much more potent in regulating presynaptic Na⁺ accumulation than HCN channels. Upon 1 s of 100 Hz firing, the presynaptic [Na⁺] increased by 5.7 ± 1.1 mm (Fig. 1B), which is larger than the overall contribution of voltage-gated Na⁺ channels and HCN channels at resting potential (Huang and Trussell, 2008, 2014). Increasing the spike frequency increased the Na⁺ transients proportionately. Spikes at 10 Hz for 60 s increased the [Na⁺] by 12.2 ± 1.7 mm, while 30 s of spikes at 20 Hz increased the [Na⁺] by 16.3 ± 2.0 mm (Fig. 1). Since the total spike number was the same in these two experiments, the difference in [Na⁺] increase likely reflects Na⁺ extrusion, presumably by Na⁺/K⁺-ATPase activity. The calyx does not fire spontaneously in brain slices; however, it fires in vivo at frequencies >70 Hz in the absence of sound and up to 350 Hz with 80 dB tones (Lorteije et al., 2009). The presynaptic [Na⁺] thus would be substantially higher in vivo than that in slice preparations, suggesting a mechanism that strongly modulates vesicle endocytosis in vivo. The presynaptic cytosolic [Na⁺] is well correlated with the firing frequency and duration (Fig. 1), therefore precisely reflects the level of vesicle exocytosis.

We showed here that intracellular Na⁺ accelerated slow and rapid endocytosis, and facilitated endocytosis overshoot. It is crucial to understand the molecular mechanism of how Na⁺ modulates endocytosis. Previous studies demonstrated that cytosolic neurotransmitter concentration regulates vesicle cycling at GABAergic terminals in hippocampal cultures (Wang et al., 2013), and the neurotransmitter concentration rapidly controls vesicular neurotransmitter contents (Hori and Takahashi, 2012; Apostolides and Trussell, 2013). Na⁺/H⁺ exchangers are expressed on synaptic vesicles (Goh et al., 2011). The elevated cytosolic Na⁺ activates vesicular Na⁺/H⁺ exchangers and regulates vesicular glutamate uptake (Huang and Trussell, 2014; Li et al., 2020), suggesting that Na⁺ may modulate the endocytosis rate via changing vesicular contents. We tested this hypothesis. However, our results showed that neither slow nor rapid modes of endocytosis were influenced by the cytosolic glutamate concentration (Fig. 5). These results are consistent with previous experiments in the calyx of Held (Hori and Takahashi, 2012; Takami et al., 2017). Isotypes of Na⁺/H⁺ exchanger, which transports Na⁺ into and H⁺ out of the cell, are also expressed on plasma membrane. By interacting plasma membrane Na⁺/H⁺ exchangers, Na⁺ may control intracellular pH, which in turn affects clathrin assembly and endocytosis (Morgan et al., 2003). In our electrophysiology and imaging recordings, the intracellular solutions contained 10 mm HEPES and were buffered to pH 7.3, which would minimize the pH change during endocytosis recordings. Moreover, by affecting the Na⁺/H⁺ exchanger, higher intracellular Na⁺ would lower intracellular pH (intracellular acidification). However, a previous study showed that a rise in intracellular alkalization facilitates endocytosis (Zhang et al., 2010), which would be opposite to the proposal if Na⁺ modulates endocytosis by controlling the activity of plasma membrane Na⁺/H⁺ exchangers. Na⁺-activated K⁺ channels are also expressed in the calyx terminal (Kaczmarek et al., 2005); however, the endocytosis was measured under Cs⁺-based intracellular solutions, which would preclude the possible contribution of K⁺ channels.

Another possible mechanism underlying the effects of Na⁺ on endocytosis is the modulation of intracellular Ca²⁺. Na⁺ has been shown to affect presynaptic Ca²⁺ homeostasis through controlling the Na⁺/Ca²⁺ exchanger activity (Kim et al., 2005), while Ca²⁺ plays specific roles in synaptic vesicle endocytosis (Leitz and Kavalali, 2016). Our results showed that changes in extracellular Na⁺ or inhibition of NCX activity did not affect the endocytosis rate (Fig. 6), and that cytosolic Na⁺ did not affect spike-evoked Ca²⁺ transients or decay, hence implying that the role of Na⁺ in accelerating endocytosis is unlikely to be mediated by controlling intracellular Ca²⁺ level and Ca²⁺-dependent pathways. Moreover, a previous study showed that subpicoampere Ca²⁺ influx has a profound effect on exocytosis (Awatramani et al., 2005), while we recently showed that Na⁺ does not affect exocytosis or neurotransmitter release probability (Li et al., 2020), indicating that changes in intracellular [Na⁺] have hardly any effect on basal Ca²⁺ level. The Ca²⁺ imaging may not be able to distinguish the Ca²⁺ change at nanodomains; however, previous studies showed that bulk Ca²⁺ would be primarily involved in endocytosis during repetitive simulations at the calyx of Held at similar ages (Yamashita et al., 2010).

The underlying mechanisms are not known. Recent studies showed that Na⁺ is involved in the activation of endocytic protein dynamin (Chappie et al., 2010) as it allosterically modulates different G-proteins (Vickery et al., 2018), suggesting a possible interaction of Na⁺ with endocytic machinery. However, the blocking of dynamin fully disrupts the endocytosis (Hosoi et al., 2009). New approaches are needed to better separate these processes experimentally.

We showed here that presynaptic cytosolic Na⁺ accelerated both slow and rapid forms of endocytosis and facilitated endocytosis overshoot. During high-frequency synaptic transmission, large numbers of synaptic vesicles are fused and Na⁺ concomitantly accumulated in the presynaptic cytosol. Na⁺ facilitates vesicle endocytosis (Figs. 1–3) and vesicular glutamate uptake (Li et al., 2020). Hence, Na⁺ works as a signal to coordinate vesicle endocytosis and vesicular glutamate uptake according to the level of exocytosis, representing a novel mechanism of vesicle exoendocytosis coupling that sustains neurotransmission at high frequency.